Status & releases

The biggest of these is a correction. A widget could not be changed after it was created unless it was a survey, and the trial handed new accounts the tightest limits in the product rather than the most generous.

Enhancements

• Edit a live widget. Every widget type now has a Build tab: the launcher label, colour, position, headline and prompt are all editable after creation. Previously only surveys could be changed, and everything else was fixed at the moment you created it.
• Choose which feedback types a widget accepts. A feedback widget can now offer any combination of bug, idea and general comment. If you only want bug reports, the other tabs disappear rather than sitting there inviting the wrong thing.
• Delete a submission. Feedback items, survey responses, annotations and board votes can be removed from the dashboard, along with any attachments.
• The trial includes everything on Pro. Twenty-five widgets, five projects, five seats, all six widget types, AI insights and follow-ups, and both add-ons. The one exception is branding removal, which stays until you are on a plan — it is the only entitlement whose expiry would show on your own public site.
• A client view for agencies. Give someone access to one project and they now see a dashboard carrying your name and logo, opening straight into their own site, with no billing, no settings and nothing selling them our plans.
• Tell us what you are setting up. New workspaces are asked whether they are for your own site or for client sites, and the first run adapts.

Fixes

• A read-only roadmap board invited visitors to "be the first to suggest one", with no way to post. It now says what it means.
• The changelog widget was missing its Build tab.
• Per-project editors could not delete a survey or annotation submission they were otherwise allowed to manage.
• Uptime on the status page counted by a window-wide estimate, which let a 30-day figure report fewer failures than the 7 days inside it. It now counts every five-minute slot per endpoint, and a missing check counts as downtime.
• Adding a monitored endpoint used to rewrite uptime history, because every past round was retroactively expected to have included it.

Two things arrived together: a way for survey respondents to answer a question you did not think to ask, and a route onto WordPress sites that does not involve editing a theme file.

Enhancements

• Survey choice questions can offer an "Other" box. Turn on allowOther for a single- or multiple-choice question and respondents get a write-in. What they type stays private to your dashboard — the public chart shows one "Other" bucket and never the text, so a poll on a public page cannot become a place where strangers publish to your visitors.
• The WordPress plugin is in the official directory. Plugins → Add New, search Feedback Magic, install, then paste a key per widget type under Settings. It carries a field for all six widget types and enqueues the hosted widget rather than bundling it, so the widget improves without a plugin update.
• Optional screening on write-ins. Where you invite free text from strangers, filterProfanity screens submissions for slurs and abuse and asks the writer to rephrase. It is off by default and deliberately so: a furious bug report is still a useful bug report.
• The survey builder shows the "Other" label, including in the preview, so you can see what respondents will see before publishing.

Fixes

• The changelog widget's RSS feed carried the old name in its title.
• Removed superseded prices from the structured data on public pages, which had outlived a pricing change.

If you point the changelog widget at one of your own elements with data-trigger, we stop drawing our floating launcher — that is the whole point. But the unread dot lived on that launcher, so choosing your own button quietly cost you the one thing that makes a changelog worth having: some indication that there is anything new to read.

We found this by running the widget in our own dashboard and noticing our "What's new" button never changed.

Enhancements

• The changelog widget now reports unread state to your page, so you can draw the indicator yourself, in your own design. Three ways, because the right one depends on how your site is built:
  • data-fm-unread="true" is set on your trigger element. Style it with a CSS ::after and write no JavaScript at all.
  • A fm:unread:<your key> event fires on window whenever the state changes, carrying detail.unread. This is the one to use if your app renders its own navigation.
  • window.FeedbackMagic.unread can be read at any time, for code that starts listening too late to catch the event.
• We deliberately do not draw into your element. It is your markup and your design system, and injecting a positioned dot into it would break layouts we cannot see. You get the state; the styling stays yours.

Fixes

• A custom trigger no longer costs the changelog its unread indicator. The two features worked correctly on their own and cancelled each other out when combined.
• Note one limit, since it is easier to plan around than debug: the data-fm-unread attribute can only be applied to elements that exist at the moment the state changes. If your navigation is rendered later by a single-page app, listen for the event instead of styling the attribute.

Feedback Magic has been live and in use for a while, but until now there was no public record of what changed and when. That starts here. Version 1.0.0 marks the line: everything below is what the product does today, and every release after this one will list what was added and what was fixed.

Enhancements

• Six widget types on every plan — feedback, surveys and CSAT, visual annotation, changelog, a public voting board, and testimonial collection. What separates the plans is how many widgets and projects you get, never which types you are allowed to use.
• Every widget also works as a link. Each one has a hosted page you can send by email, put behind a QR code on a receipt, paste into a support reply or share on social. No website required.
• Channel attribution — responses are tagged by how they arrived (embed, link, email, QR, SMS, social, support or print) with a 30-day breakdown in the dashboard, so you can see which route people actually used.
• AI triage on every submission — each item is categorised, tagged and given a severity automatically, with weekly digests and ask-your-feedback on Pro.
• Pip, the built-in assistant — answers questions about the product on every plan including during the trial, and can design a widget for you from a sentence as an optional add-on.
• Flat pricing. Responses, voters, page views and reviewers are unlimited and never metered. The bill does not grow when your feedback programme works.
• Placement that fits your site — four launcher positions, or point the widget at one of your own buttons with data-trigger so it lives in your nav, your footer or your help menu instead of floating in a corner.
• Visitors can be allowed to dismiss a widget for their session, off by default and enabled per widget.
• Install however you build — a one-line script tag, the feedback-magic npm package with a React component and a custom element, a WordPress plugin, or an MCP server so AI editors can create widgets for you.
• Per-project roles — give a client viewer or editor access to a single project without making them a member of your workspace.
• Outbound integrations — push submissions to GitHub Issues, Jira, Zendesk, Freshdesk, Freshservice, Twilio or your own webhook.

Fixes

Nothing to report — this is the baseline release. Fixes will be listed here from the next release onward.